MedDelinea: Scalable and Efficient Medical Image Segmentation via Controllable Diffusion Transformers

Gayatri Deshmukh, Onkar Kishor Susladkar, Debesh Jha, Elif Keles, Halil Ertugrul Aktas, Alpay Medetalibeyoglu, Daniela P. Ladner, Amir A. Borhani, Gorkem Durak, Ulas Bagci
Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, PMLR 301:327-345, 2026.

Abstract

We introduce MedDelinea, a novel medical image segmentation architecture that leverages a controllable module, drawing inspiration from ControlNet, within the Diffusion Transformers (DiT) framework. By doing so, we effectively address three key challenges inherent to segmentation tasks: (1) limited availability of labeled data, (2) variability in image modalities, and (3) the need for precise boundary delineation. MedDelinea is pre-trained on a large-scale medical dataset, thereby mitigating overfitting risks and enabling efficient transfer across diverse imaging scenarios with minimal fine-tuning requirements. The modular design of MedDelinea facilitates scalable and efficient computation, while maintaining high-quality segmentation performance in both supervised and zero-shot settings. Through extensive empirical evaluations on multiple datasets, we demonstrate that MedDelinea outperforms existing state-of-the-art segmentation approaches, showcasing its potential for robust and accurate medical image analysis

Cite this Paper

BibTeX
@InProceedings{pmlr-v301-deshmukh26a, title = {MedDelinea: Scalable and Efficient Medical Image Segmentation via Controllable Diffusion Transformers}, author = {Deshmukh, Gayatri and Susladkar, Onkar Kishor and Jha, Debesh and Keles, Elif and Aktas, Halil Ertugrul and Medetalibeyoglu, Alpay and Ladner, Daniela P. and Borhani, Amir A. and Durak, Gorkem and Bagci, Ulas}, booktitle = {Proceedings of The 8th International Conference on Medical Imaging with Deep Learning}, pages = {327--345}, year = {2026}, editor = {Tasdizen, Tolga and Elhabian, Shireen and Summers, Ronald and Chen, Chen and Koch, Lisa and Zhuang, Yan}, volume = {301}, series = {Proceedings of Machine Learning Research}, month = {09--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v301/main/assets/deshmukh26a/deshmukh26a.pdf}, url = {https://proceedings.mlr.press/v301/deshmukh26a.html}, abstract = {We introduce MedDelinea, a novel medical image segmentation architecture that leverages a controllable module, drawing inspiration from ControlNet, within the Diffusion Transformers (DiT) framework. By doing so, we effectively address three key challenges inherent to segmentation tasks: (1) limited availability of labeled data, (2) variability in image modalities, and (3) the need for precise boundary delineation. MedDelinea is pre-trained on a large-scale medical dataset, thereby mitigating overfitting risks and enabling efficient transfer across diverse imaging scenarios with minimal fine-tuning requirements. The modular design of MedDelinea facilitates scalable and efficient computation, while maintaining high-quality segmentation performance in both supervised and zero-shot settings. Through extensive empirical evaluations on multiple datasets, we demonstrate that MedDelinea outperforms existing state-of-the-art segmentation approaches, showcasing its potential for robust and accurate medical image analysis} }
Endnote
%0 Conference Paper %T MedDelinea: Scalable and Efficient Medical Image Segmentation via Controllable Diffusion Transformers %A Gayatri Deshmukh %A Onkar Kishor Susladkar %A Debesh Jha %A Elif Keles %A Halil Ertugrul Aktas %A Alpay Medetalibeyoglu %A Daniela P. Ladner %A Amir A. Borhani %A Gorkem Durak %A Ulas Bagci %B Proceedings of The 8th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Tolga Tasdizen %E Shireen Elhabian %E Ronald Summers %E Chen Chen %E Lisa Koch %E Yan Zhuang %F pmlr-v301-deshmukh26a %I PMLR %P 327--345 %U https://proceedings.mlr.press/v301/deshmukh26a.html %V 301 %X We introduce MedDelinea, a novel medical image segmentation architecture that leverages a controllable module, drawing inspiration from ControlNet, within the Diffusion Transformers (DiT) framework. By doing so, we effectively address three key challenges inherent to segmentation tasks: (1) limited availability of labeled data, (2) variability in image modalities, and (3) the need for precise boundary delineation. MedDelinea is pre-trained on a large-scale medical dataset, thereby mitigating overfitting risks and enabling efficient transfer across diverse imaging scenarios with minimal fine-tuning requirements. The modular design of MedDelinea facilitates scalable and efficient computation, while maintaining high-quality segmentation performance in both supervised and zero-shot settings. Through extensive empirical evaluations on multiple datasets, we demonstrate that MedDelinea outperforms existing state-of-the-art segmentation approaches, showcasing its potential for robust and accurate medical image analysis
APA
Deshmukh, G., Susladkar, O.K., Jha, D., Keles, E., Aktas, H.E., Medetalibeyoglu, A., Ladner, D.P., Borhani, A.A., Durak, G. & Bagci, U.. (2026). MedDelinea: Scalable and Efficient Medical Image Segmentation via Controllable Diffusion Transformers. Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 301:327-345 Available from https://proceedings.mlr.press/v301/deshmukh26a.html.

Related Material